How to

These guides walk through common AI Search use cases, from no-code setups to full agent applications. Use the difficulty column to find the right place to start.

Guide	Difficulty	What you build
Add search to your website	Beginner	Add a search bar, chat bubble, and search modal to a site with the prebuilt UI snippets.
Connect your AI Search to an MCP client	Beginner	Expose your indexed content as a search tool for any MCP client or AI assistant, with no code.
Create a simple search engine	Beginner	Query an instance from a Worker using the search binding.
Multitenancy	Intermediate	Keep each tenant's data separate with a per-tenant instance or a shared instance and metadata filtering.
Search multiple sources at once	Intermediate	Search a shared knowledge base and a tenant-specific one together in a single query.
Fetch and index single web pages	Intermediate	Use Browser Run to fetch a page's rendered HTML, index it, and search it from one Worker.
Bring your own generation model	Intermediate	Use AI Search for retrieval and an external model, such as OpenAI, for generation.
Show source citations in responses	Intermediate	Return AI-generated answers with citations to the source documents that informed them.
Human-in-the-loop knowledge base updates	Advanced	Build an agent that proposes new documents and pauses for human approval, with rollback.